Citation Rates for Speeding Stops

Scatter Jitter

Photo by Nomadsoul1 on Freepik

If you don’t want a speeding ticket, don’t speed… — Ainsley Earhardt

From the Stanford Open Policing Project: “On a typical day in the United States, police officers make more than 50,000 traffic stops. Our team is gathering, analyzing, and releasing records from millions of traffic stops by law enforcement agencies across the country. Our goal is to help researchers, journalists, and policymakers investigate and improve interactions between police and the public.”

This example is based on summary data with the following fields, types, and definitions:

location: character, County/District location for each incidence
state: character, State for each incidence
driver_race: character, driver’s race
stops_per_year: double, number of stops per year
stop_rate: double, as percentage
search_rate: double, as percentage
consent_search_rate: double, as percentage
arrest_rate: double, as percentage
citation_rate_speeding_stops: double, as percentage
hit_rate: double, the proportion of searches that successfully turn up contraband
inferred_threshold: double, based off the threshold test

Ingest data

summary data

df <- read.csv("archetypes/citation-rates-for-speeding-stops/citation-rates-for-speeding-stops.csv", header = TRUE, stringsAsFactors = FALSE)
df

Wrangle data

summarize by state and race

byStateAndRace <- df %>%
  group_by(location, state, driver_race) %>%
  summarize(meanRate = mean(citation_rate_speeding_stops))

byStateAndRace

# unique(byStateAndRace$driver_race)

Plot Scatter-Jitter

facet by state, shape by race

theme_opts <- theme(
    text = element_text(family = "inconsolata", size = 10), 
    axis.text.x = element_blank(),
    plot.title = element_text(color = "black", size = 16, face = "bold"),
    plot.subtitle = element_text(color = "black", size = 12),
    plot.caption = element_text(color = "#555555", size = 10),
    panel.border = element_rect(colour = "black", fill=NA, size=1),
    panel.background = element_rect(colour = NA, fill="#F9F9F9", size=1),
    panel.spacing.x = unit(1, "lines"),
    panel.spacing.y = unit(1, "lines"),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),
    strip.text = element_text(size = 11, face = "bold"),
    strip.background = element_blank(),
    legend.position='top',
    legend.title = element_blank()
)

v1 <- ggplot(byStateAndRace, aes(x = location, y = meanRate)) +
  geom_point(aes(shape = factor(driver_race)), color = "black", size = 1, alpha = 0.6) +
  scale_shape_manual(values=c(1, 3, 8)) +
  coord_cartesian(ylim = c(0, 1), expand = TRUE) +
  facet_grid(.~state, scales = "free") +
  labs(title =  "Citation rates for speeding stops",
       subtitle = "as reported by counties or districts",
       x = NULL,
       y = "citation rate",
       caption = "Source: arXiv:1706.05678") +
  theme_minimal() +
  theme_opts

girafe(ggobj = v1, width_svg = 1280/72, height_svg = 720/72,
       options = list(opts_sizing(rescale = TRUE, width = 1.0))
)

Plot

color with split complements palette

When selecting a color palette there can be subtle biases that influence perception. In this example, using race as a color variable, we have a factor of three categories: black, hispanic, and white. This provides a natural fit for evaluating schemes based on color theory. One option is Split Complements, the use a color and the two adjacent tertiary colors of its complement. But take notice that two of the colors are more similar than the third.

byStateAndRace_complete <- byStateAndRace[complete.cases(byStateAndRace), ]

race_palette_sc <- c(
  "Black" = "#453C97",
  "Hispanic" = "#009899",
  "White" = "#FBAA19"
)

v2 <- ggplot(byStateAndRace_complete, aes(x = location, y = meanRate)) +
  geom_point(aes(colour = factor(driver_race), shape = factor(driver_race)), size = 1, alpha = 0.6) +
  coord_cartesian(ylim = c(0, 1), expand = TRUE) +
  scale_color_manual(values = race_palette_sc) +
  scale_shape_manual(values=c(1, 3, 8)) +
  facet_grid(.~state, scales = "free") +
  labs(title =  "Citation rates for speeding stops",
       subtitle = "as reported by counties or districts",
       x = NULL,
       y = "citation rate",
       caption = "Source: arXiv:1706.05678") +
  theme_minimal() +
  theme_opts

girafe(ggobj = v2, width_svg = 1280/72, height_svg = 720/72,
       options = list(opts_sizing(rescale = TRUE, width = 1.0))
)

Plot

color with triadic palette

In contrast to split the split complements palette, triadic color schemes use three evenly spaced colors on the color wheel. Therefore there is no bias to any assignment of color to the race factors.

race_palette_triadic <- c(
  "Black" = "#CF1876",
  "Hispanic" = "#009899",
  "White" = "#FED502"
)

v3 <- ggplot(byStateAndRace_complete, aes(x = location, y = meanRate)) +
  geom_point(aes(colour = factor(driver_race), shape = factor(driver_race)), size = 1, alpha = 0.6) +
  coord_cartesian(ylim = c(0, 1), expand = TRUE) +
  scale_color_manual(values = race_palette_triadic) +
  scale_shape_manual(values=c(1, 3, 8)) +
  facet_grid(.~state, scales = "free") +
  labs(title =  "Citation rates for speeding stops",
       subtitle = "as reported by counties or districts",
       x = NULL,
       y = "citation rate",
       caption = "Source: arXiv:1706.05678") +
  theme_minimal() +
  theme_opts

girafe(ggobj = v3, width_svg = 1280/72, height_svg = 720/72,
       options = list(opts_sizing(rescale = TRUE, width = 1.0))
)

References

citations for narrative and data sources

Narrative sources: A large-scale analysis of racial disparities in police stops across the United States, GO
Data sources: The Stanford Open Policing Project TidyTuesday, GO